784 Chem. Mater. 1994, 6, 784-187

Encapsulated Nanoparticles of Iron Metal K. J. Klabunde,' D. Zhang, G. N. Glavee, and C. M. Sorensen

Departments of Chemistry and Physics, Kansas State University, Manhattan, Kansas 66506

G. C. Hadjipanayis Department of Physics and Astronomy, University of Delaware, Newark, Delaware 19716

Received January 3, 1994. Revised Manuscript Received March 14, 1994"

Metastable iron-magnesium particles as intimate mixtures were prepd. by co-depositing Fe and Mg vapors with a hydrocarbon diluent a t 77 K, followed by warming to room temperature. Upon heat treatment phase segregation was accomplished with reasonable control. Core-shell particles (Fe as core) were formed, where Fe particles were encapsulated, and with appropriate heat treatment could be completely protected from surface oxidation. Crystallite sizes of the Fe core ranged from 3 to 26 nm depending on the heat treatment procedure. The magnetic properties of these encapsulated particles indicated extremely soft ferromagnetism with very low coercivities and with little size dependency. These results coupled with earlier findings suggest that coercivity can be controlled (dominated) by surface coating in such small particles.

Introduction

Although numerous studies of the magnetic properties of nanoparticles of iron metal have been reported, the question as to whether surface contamination by thin oxide (FezOs/FesOr) coatings was present has not been satis- factorily resolved.'-9 From the point of view of a chemist, the synthesis of gram quantities of pure, oxide-free nanoparticles of any oxophilic metal is a genuine night- mare. Even working under the best anaerobic conditions (parts per million oxygen in argon), surface oxidation cannot be avoided with such high surface area, reactive materials. And since it is becoming increasingly clear that even the smallest contamination by magnetic oxides (e.g., FesO4) can have dramatic effects on coercivity and saturation magnetization values of nanoparticles,lOJ1 there is a need to develop a means of encapsulating the Fe particles in a protective coating while still retaining the morphology of a free flowing powder.

Results and Discussion

The approach we have used is to create metastable alloy particles of the thermodynamically immiscible pairs Fe- Li and Fe-Mg by codepositing the vapors of Fe and Mg (or Li) into a freezing hydrocarbon matrix at 77 K (the

Abstract published in Advance ACS Abstracts, May 1, 1994. (1) Matijevic, E.,Mater.Res. Bull. 1989,14,18-20. Annu.Rev. Mater.

(2) Ozaki, M. Mater. Res. Bull. 1989, 14, 35-40. (3) Tasaki, A.; Takao, M.; Tokunaga, H. Jpn. J . Appl. Phys. 1974,13,

SOC. 1985, 15, 483.

271. -. -. (4) Christensen, P. H.; Morup, S.; Niemantaverdrit, J. W. J. Phys.

(5) Jen,S.U.;Lee,C.Y.;Yao,Y.D.;Lee,K.C.J.Magn.Magn.Mater. Chem. 1985,89, 4898.

1991, 82, 96. (6) Chien, C. L. J. Appl. Phys. 1991, 69, 5269. (7) Gong, W.; Li, H.; Zhao, Z.; Chen, J. J. Appl. Phys. 1991,69, 5119. (8) Schabes, M. E. J. Appl. Phys. 1991,69,4487. (9) Du, Y. W.; Wu, J.; Lu, H. X.; Wang, T. X.; Qin, Z. Q.; Tang, H.;

Walker, J. C. J . Appl. Phys. 1987, 61, 3314. (10) Gangopadhyay, S.; Hadjipanayis, G. C.; Dale, E. B.; Sorensen, C.

M.; Klabunde, K. J.; Papaefthymiou, V.; Kostikas, A. Phys. Rev. B 1992, A.5 977A

1991, 82, 96. (6) Chien, C. L. J. Appl. Phys. 1991, 69, 5269. (7) Gong, W.; Li, H.; Zhao, Z.; Chen, J. J. Appl. Phys. 1991,69, 5119. (8) Schabes, M. E. J. Appl. Phys. 1991,69,4487. (9) Du, Y. W.; Wu, J.; Lu, H. X.; Wang, T. X.; Qin, Z. Q.; Tang, H.;

Walker, J. C. J . Appl. Phys. 1987, 61, 3314. (10) Gangopadhyay, S.; Hadjipanayis, G. C.; Dale, E. B.; Sorensen, C.

M.; Klabunde, K. J.; Papaefthymiou, V.; Kostikas, A. Phys. Rev. B 1992, A.5 977A --, I, . ".

(11) Gangopadhyay, S.; Hadjipanayis, G. C.; Sorensen, C. M.; Klabunde, K. J. Phys. Rev. Lett., submitted.

procedure has been described in detail earlier).12 Initially a frozen matrix containing Fe and Mg atoms/clusters is formed which upon slight warming allows accretion into larger Fe-Mg alloy-like particles.13a In this way the growth of Fe-Mg particles at low temperature is controlled by kinetic parameters, such as metal concentration, rate of temperature increase, and mobility of metal atoms and the smaller clusters. As the growth continues the clusters/ particles become less mobile and growth eventually stops. In this way individual particles (not films) are produced by this matrix method.lsb Although this process is somewhat analogous to earlier reports of the production of granular film composites of Fe-Ag,14 Fe-Mg,16 and Fe- Hg,l8 product morphology is completely different due to the hydrocarbon matrix diluent, and free-flowing powders are obtained.

Since Fe-Mg and Fe-Li metals are immiscible under thermodynamic control,17 we reasoned that by careful heat processing of these metastable alloy particles, phase separation could be achieved to yield a core-shell structure where Mg(Li) would form a coating around the nano- particles of Fe (Figure 1). Indeed, this approach works very well, and Fe atoms/clusters migrate to the center of the particles. Although it could not be predicted a priori that this type of core-shell structure would be formed, we considered that under heating thermodynamic factors would be controlling, in that Fe-Fe bonds are stronger than Mg-Mg or Fe-Mg bonds and that this should

(12) Glavee et al. (Glavee, G. N.; Kernizan, C. F.; Klabunde, K. J.; Sorensen, C. M.; Hadjipanayis, G. C. Chem. Mater. 1991,3,967) reported on the synthesis of Fe-Li particlee by a similar method.

(13) (a) Klabunde, K. J. Chemistry of Free Atoms and Particles; Academic Press: New York, 1980; p 114. (b) For further discussion of the growth process see: Wang, Y.; Li, Y. X.; Klabunde, K. J. In Selectivity in Catalysis; ACS Symposium Series No. 517; Chap. 10, Davis, M. E., Suib, S. L., Ed.; American Chemical Society: Washington, DC, 1993; pp

(14) Tako, M.; Senno, H. J. Magn. Magn. Mater. 1983,31-34,949. (15) Kawaguchi, K.; Yamamoto, R.; Hosoito, N.; Shinjo, T.; Takada,

(16) Bean, C. P.; Jacobs, I. S. J. Appl. Phys. 1956,1448, 27(12). (17) Kubaschewski, 0. Iron-Binary Phase Diagrams; Springer-

136-155.

T. J. Phys. SOC. Jpn. 1986,55, 2375.

Verlag: Berlin, 1982; 59.

0897-4756/94/2806-0784$04.50/0 0 1994 American Chemical Society

Encapsulated Nanoparticles of Iron Metal Chem. Mater., Vol. 6, No. 6, 1994 786

1 warm

aMg Fe

Figure 1. Schematic of the formation of Fe-Mg alloy nanw particles followed by phase segregation.

encourage the formationof acoatingof Mg on anFe core.lS This has been found to be the case with both Fe-Mg and Fe-Li.12 However, with FeAg alloy particles, the situation isnotsosimple,andamixtureofFe.,Ag.,andunprotected Fe-Ag particles was formed.*@

One of the advantages of this approach where Mg (or Li) is thesecondcomponent is that it behavesasasacrificial metal. That is, if traces of oxygen enter any apparatus where the nanoparticles are being manipulated, MgO is formed (itself a good protective coating).

Heat treatment of the as-prepared samples leads to increasing a-Fe particle sizes as seen in Figure 2. Mass- bauer spectroscopy after heat treatment and subsequent exposure to air showed the presence of iron oxide for heat treatments at temperatures T 5 200 oC.m.*' On the other hand, for T > 250 'C no oxide was detected. This implies the magnesium formed an encapsulation effective for protecting the iron only when heat treatment was greater than 250 "C, which corresponds to a-Fe sizes greater than 7 nm.** The Fe-core/Mg-shell structure, which we des- ignate as [MglFe, is illustrated by TEM photos of hea t treated, passivated samples (Figure 3 and 4). The darker areas are due to a-Fe particle cores, while the lighter outer shells are [MgOlMg (MgO oxide coated Mg metal). We belieue the a-Fe cores are single crystals since TEM measurements and XRD crystallite sizes calculated by

(18) (a) A popular expression for the spontaneous ordering of surface adsorb- is 'molecular self-assembly". In our Fe-Li and Fe-Mg metastable alloy particles, there is 'atomic wlf--assembly" into coreahell structures when experimental heating parameters are mrrect. (b) A reviewer hassuggested thstthetempe*8tlll~v8listion afthesurfaceenergy of iron vs magnesium may be important here. (19) Easom. K.; Klsbunde. K. J.; Sorensen, C. M.; Hadjipanaysis, G.

C. Polyhedron, in press. (20) Samples were placed in g l w vials under Ar, the screw cap WBLI

l o w e d . and air was allowed to alowly leak in overnight (21) Samples were placed in waled glaaa vials under Ar and healed

at the desired temperature for 1 h. (22) We believe that our individual particlea phaae m a t e i n d e

pendent of each other. Upon prolonged heating at 430 O C or above, the p+clea begin to neck w e t h e r and to amalgamate. and particles along wrth a-Fe crystallite aizes gmw further. This is undesirable since we want to preserve small particle size. With such a system, we cannot be awe if individual particles completely phase segregate before the onset of interpMiclesmdgametion. Nonetheless. therearesimilaritieato~~lier studies of precipitation of crystallites in bulk allays such as C d u (see: Beeker. J. J.; Trow. A.I.M.E. 1957,209,59-63 and discussion in Cullity (ref23pp418-421). Thus.itisdiffieulttobesureifthevolumefraction of 0-Fe is affecting our results 8s was the case in the Becker work.

.

mL 0 Heat TresImM Tempr*lu&O

Figure 2. a-Fe crystallite aizes of Fe-Mg powders (formed in cold pentane). Heattreated at different temperatures for 60 min.

3 1

I 200 nm I Figure 3. TEM micrograph of pessivated FeMg powders heat p r d at 180 "C.

line broadening (Scherrer equation) are in good agree- ment.

Magnetizationdatawerecollected foraseriesofsamples. For fresh, unheat-treated, unpassified Fe-Mg (47 atomic ratio) alloy particles, a plot of magnetization (u, emu/g iron) vs applied field is shown in Figure 5. The saturation magnetization u,inferred from the high-field limitof these

786 Chem. Mater., Vol. 6, No. 6,1994

t

Klabunde et al.

Figure4. TEI processed at 4:

I I micrograph of pnsnivated FeMg powders heat O C .

2Ml nm

“i 0 T= IOK AT=IMK .T=3WK

I 0 10 20 30 40 SO 60

II(K0C)

Figure 5. Magnetization curves of a fresh FeMg eample with a Fe:Mg atomic ratio of 41.

plots is much lower than for bulk a-Fe (220 emu/g at 300 K). However, upon heat treatment and a-Fe crystallite growth, u.changed dramatically (Figure0 andapproaches the bulk value as the a-Fe particle diameter increases.

Further magnetic studies for the well encapsulated, d > 7 nm samples showed superparamagnetic behavior as identified by u vs HIT curves after a small ferromagnetic fraction was subtracted. This ferromagnetic fraction increased with heat treatment, hence average size, until a t d = 20 nm, the whole sample was ferromagnetic. Coercivity measurements as a function of size are shown in Figure 7, where we see small values independent of size.

0 T= IOK A T-ISOK

T = W K

0 5 10 I S 20 P F ~ Crymlllle Skebm)

Fqnre6. Saturationmagnetizationvalues(o,emdgramofiron) ofFeMgparticles (FeMg= 47:fieldstrength55 0000e).Smaller particles contained some iron oxide. Note that in the 1C-20-nm range near ‘bulklike” n. values were observed.

.

A T = M K T = m K

A

. a X

Figure 7. Coercivities of F e M g particles (Fe:Mg = 47) with different a-Fe crystallite si=. Error estimates are *IO%.

These nonzero H. values must be due to the small ferromagnetic fraction, but the behavior of the H, is contrary to an increase of H. with increasing size expected for particles smaller than the single domain size. To explain this, we speculate that the average size increases as a result of increasing ferromagnetic fraction, but the average size of this fraction alone does not change significantly with heat treatment. We remark that the calculated superparamagnetic size for spherical iron particles with bulk crystalline anisotropy is d = 13 nm in accord with the size range of our a-Fe crystallities.B.a

We believe these [MglFe particles represent the first examples of free-flowing powder samples of nanoe.de a-Fe that do not have surface contamination by magnetic iron oxide but instead are coated by a nonmagnetic protective shell. And since we have shown that iron oxide coatings can have dramatic effects on H, in such small particles,’O it is obvious that these [MglFe particles are undergoing magnetic domain spin flips that are controlled by a completely different process. Thus, there are no magnetic spins in the shell material that can pin the spins of the

(23) Cullity, B. D. Introduction to M e t i c Materials; Addision- (24) Glsvee, G. N.; FA”. K.; Klabunde. K. J.; Soremen, C. M.;

Wesley: Reading. MA, 1972; p 386. Hadjipanayis. G. C. Chem. Mater. 1992,4, 1380.

Encapsulated Nanoparticles of Iron Metal

a-Fe core. It appears that the Mg coating has little effect on inhibiting a flip of the a-Fe magnetic moment; there is no induced exchange anisotropy.

An important conclusion is that surface coatings in such small particles can dominate the magnetic memory property H,. Thus, in the absence of surface contaminants of magnetic iron oxides and instead coated with metallic Mg, nanoscale a-Fe particles exhibit a very, soft magnetic behavior. Our results suggest that it is the surface coating on a-Fe that dictates size dependent behavior in this property H, of fine particles.

One final point is that these heat processed particles are completely air stable, even for months at a time. The outer Mg coating is itself protected by a MgO thin coating.

Controlling coercivities and magnetization values in nanoparticles now seems more likely, and more will be learned about surface chemical effects as more core-shell particles are prepared and studied.

We are currently carrying out further investigations, for example with [MglFe, [LilFe, [FeaOrlFe, [AglFe, [InlFe, [BilFe, and [FeSlFe systems.

Chem. Mater., Vol. 6, No. 6, 1994 787

The metal vapor reactor26.28 was set up with two W-Al203 crucibles, one for Fe and the other for Mg vaporization. After pumpdown and cooling the reactor to 77 K, 40-50 mL of pentane was deposited on the inner walls. The vaporization of iron was initiated after a steady vaporization of Mg was achieved (pentane codeposition continued at a rate of 2-3 mL/min). Typical vaporization rates were about 30 mg/min (6 X l(r mol/min) and 19 mg/min (8 X lo-' mol/min) for Fe and Mg, respectively. After about 50 min the metal evaporators were shut off and an additional 30-40 mL of pentane was deposited. The reactor was isolated from the pumping system, the liquid Nz dewar removed, and the reactor allowed to warm to room temperature. The liquid pentane was removed by reapplying vacuum. Then the reactor was back filled with Ar, and the chamber removed under a flow of Ar and quickly covered and transferred to the inert-atmosphere box.

A series of Fe-Mg composites were prepared with atomic ratios of Fe/Mg of 1/2,4/7, 1/1, and 1/3. Preparation of pure Fe and pure Mg powders was carried out similarly. In all cases about 80 % of the metals vaporized were recovered as black, free flowing powders.

For further treatments, samples of 70-100 mg were transferred into Pyrex glass tubes in an inert (Ar filled) atmosphere box. The tubes were temporarily sealed, removed, and then flame sealed. Heat treatments were carried out by wrapping each sample in heating tape and heating to the desired temperature for specified times. For temperatures below 400 "C, 1 h of heating was sufficient to achieve a stable structure, and longer times had little effect. After heat treatment the samples were opened in the inert-atmosphere box, and samples prepared for XRD, Mossbauer, SQUID, and other analyses.12*u

Acknowledgment. The support of the National Sci- ence Foundation is acknowledged with gratitude.

Experimental Section General procedures for the preparation and characterization

of bimetallic, metastable fine particles have been described before. ~4

For the Fe-Mg combination, the metals used, and synthetic isolation procedures were as follows:

Metals: Iron metal chips, 99.99+ % (Fischer Scientific); pieces from a magnesium metal rod 99.9+ % purity (Aldrich Chem. Co.); pentane, HPLC grade (Fischer Scientific).

The pentane was dried by refluxing over CaH2 under a flow of Nz for at least 6 h. About 200 mL was freshly distilled and freeze-thaw degassed.

~~

(25) Klabunde, K. J.; Timms, P. L.; Skell, P. S.; Ittel, S. Inorg. Synth.

(26) Li, Y. X.; Zhang, Y. F.; Klabunde, K. J. Langmuir 1988,4,385. 1979, 19, 59.